Development and characterization of a glycine biosensor system for fine-tuned metabolic regulation in Escherichia coli

Abstract

Background: In vivo biosensors have a wide range of applications, ranging from the detection of metabolites to the regulation of metabolic networks, providing versatile tools for synthetic biology and metabolic engineering. However, in view of the vast array of metabolite molecules, the existing number and performance of biosensors is far from sufficient, limiting their potential applications in metabolic engineering. Therefore, we developed the synthetic glycine-ON and -OFF riboswitches for metabolic regulation and directed evolution of enzyme in Escherichia coli.

Results: The results showed that a synthetic glycine-OFF riboswitch (glyOFF6) and an increased-detection-range synthetic glycine-ON riboswitch (glyON14) were successfully screened from a library based on the Bacillus subtilis glycine riboswitch using fluorescence-activated cell sorting (FACS) and tetA-based dual genetic selection. The two synthetic glycine riboswitches were successfully used in tunable regulation of lactate synthesis, dynamic regulation of serine synthesis and directed evolution of alanine-glyoxylate aminotransferase in Escherichia coli, respectively. Mutants AGXT22 and AGXT26 of alanine-glyoxylate aminotransferase with an increase of 58% and 73% enzyme activity were obtained by using a high-throughput screening platform based on the synthetic glycine-OFF riboswitch, and successfully used to increase the 5-aminolevulinic acid yield of engineered Escherichia coli.

Conclusions: A synthetic glycine-OFF riboswitch and an increased-detection-range synthetic glycine-ON riboswitch were successfully designed and screened. The developed riboswitches showed broad application in tunable regulation, dynamic regulation and directed evolution of enzyme in E. coli.

Keywords: Biosensor; Directed evolution; Escherichia coli; Glycine riboswitch; Metabolic regulation.

Fig. 1

Relative GFP expression (GFP/OD₆₀₀) of different glycine riboswithches expressed in MG1655 strain. The relative GFP expressions were measured in M9 medium supplemented with 0, 7, 13, 27, 53, 80, and 107 mM glycine, and MG1655 was cultured in M9 medium without ampicillin. All data are the average values of three independent experiments

Fig. 2

Screening workflow of synthetic glycine-ON and -OFF riboswitches

Fig. 3

Selection and characterization of glycine-ON and -OFF switches from mutant libraries. a ON/OFF ratio of relative fluorescence intensity in the wild-type glycine riboswitch and the selected glycine-ON riboswitches in M9 medium with or without 80 mM glycine in 96-deep-well plates. b ON/OFF ratio of relative fluorescence intensity in the wild-type glycine riboswitch and the selected glycine-OFF riboswitches in M9 with or without  80 mM glycine in 96-deep-well plates. The yellow column indicates the control strain. c GFP/OD₆₀₀ ratio of MG1655-BSTG, G1655-BS-ON14, and MG1655-BS-OFF6. Strains were cultured in M9 medium supplemented with 0, 7, 13, 27, 80, and 107 mM glycine. All the data are the average values of three independent experiments. d Secondary structure of the wild-type B. subtilis glycine riboswitch. e Mutations in the glycine-OFF riboswitch glyOFF6 affecting the secondary structure. f Mutations in the glycine-ON riboswitch glyON14 affecting the secondary structure. Base-pairing stems were labeled as the P0/kink-turn motif, P1, P2, and P3 with subsections labeled ‘a’ and ‘b’, and junction structures were labeled J1, J2 and JD with subsections labeled ‘*’. Mutant sites were highlighted in green nucleotides and green boxes with mutation information, red shading highlights nucleotides that base pair to form the transcription terminator stem when the riboswitch is in the ‘OFF’ conformation. Riboswitches’ structures and its mutations were analyzed by RNAflod and VARNA 3–93

Fig. 4

Tunable regulation of lactate synthesis using the evolved glycine-ON riboswitch glyON14 in E. coli. a Schematic representation of the metabolic pathways involved in lactate production from glucose in E. coli. The expression of ldhA, encoding lactate dehydrogenase, was dynamically controlled under the synthetic glycine-ON riboswitch glyON14 in the high-copy-number plasmid pUC18. GLU, glucose; PGA, 3-phosphoglycerate; PEP, phosphoenolpyruvate; ISO, isocitrate; AKG, α-ketoglutaric acid; FUM, fumaric acid; TCA cycle: tricarboxylic acid cycle; PYR, pyruvate; LAC, lactate; Ac-CoA, acetyl coenzyme A; Amp, ampicillin; r, resistance; glycine-ON riboswitch glyON14 in the dashed line box; ori, replicon. b Lactate production of engineered strain W105-18A and the control strain W105-18P in M9 medium supplemented with 0, 7, 13, 27, and 80 mM glycine. c Maximal lactate concentration of W105-18A and W105-15A in M9 medium supplemented with 0, 7, 13, 27, and 80 mM glycine. Error bars represent the standard deviations of triplicate samples.**P < 0.01; *P < 0.05

Fig. 5

Dynamic metabolic regulation for improved serine production in E. coli using the glycine-OFF riboswitch glyOFF6. a Schematic representation of the metabolic pathways involved in serine production from glucose in recombinant E. coli. The expression of glyA, encoding serine hydroxymethyltransferase, was dynamically regulated under the synthetic glycine-OFF riboswitch glyOFF6 in plasmid pUC18, co-expressed with pPK10 to produce serine in E. coli. The genes sdaA, encoding L-serine deaminase I; sdaB, encoding L-serine deaminase II; and tdcG, encoding L-serine deaminase III were deleted, serA^Δ197, (encoding a truncated 3-phosphoglycerate dehydrogenase gene from C. glutamicum), serC (encoding phosphoserine aminotransferase), serB (encoding phosphoserine phosphatase) and pgk (encoding phosphoglycerate kinase) were upregulated by overexpression using the plasmid pPK10. GLU, glucose; PGA, 3-phosphoglycerate; PEP, phosphoenolpyruvate; ISO, isocitrate; AKG, α-ketoglutaric acid; FUM, fumaric acid; OAA, oxaloacetic acid; TCA cycle: tricarboxylic acid cycle; PGA, 3-phosphoglycerate; PEP, phosphoenolpyruvate; PYR, pyruvate; Ac-CoA, acetyl coenzyme A; PAP, 3-phosphopyruvate; S3P, 3-phosphoserine; MTPG, methylene-tetrahydropteroyl polyglutamate; AMHDP, s-amino-methyldihydrolipoyl protein; GLY, glycine; SER, serine; Amp: ampicillin, Cr: chloramphenicol resistance gene; ori, replicon; glycine-OFF riboswitch glyOFF6 in the solid line box; b OD₆₀₀, glucose consumption, and serine production of engineered strain E4GS and control strain E4GD in M9Y medium under aerobic conditions with an initial OD₆₀₀ of 0.05. (Black solid circle: E4GD OD₆₀₀; Black solid square: E4GD OD₆₀₀; Blue falling triangle: E4GD glucose consumption; Blue positive triangle: E4GS glucose consumption; Red diagonal triangle: E4GD serine production; Red rhombus: E4GS serine production) Error bars represent the standard deviations of triplicate samples

Fig. 6

Synthetic glycine-OFF riboswitch used for the directed evolution of alanine-glyoxylate aminotransferase in E. coli. a A mutant of AGXT with higher enzyme activity was screened using a high-throughput screening platform based on the synthetic glycine-OFF riboswitch. The genes agxtM, aceA and hemA were co-expressed from two plasmids to syntheze 5-ALA in E. coli. The glyoxylate pathway is highlighted in red. b The library of agxT mutants was constructed using error-prone PCR (Additional file 1: Table S6). The library plasmids and puc18-GlyOFF6-aceA plasmid were co-expressed in E. coli MG1655. c All clones from LB plates were transferred to M9 medium, and selected in fresh M9 medium supplemented with 100 μg mL⁻¹ ampicillin and 90 μM Ni²⁺ to remove the strains with lower levels of AGXT enzyme activity. AGXT mutants with higher activity were enriched in three cycles, and the plasmids from each cycle were extracted and transferred to the original strain. The percentage of beneficial mutations and enzyme activity were also determined. The relevant genes include aceA, encoding isocitrate lyase; agxTM, encoding mutant glyoxylate transaminase; hemA, encoding 5-aminolevulinic acid synthase; tetA, tetracycline resistance gene; gfp, encoding green fluorescent protein. Ac-CoA, acetyl coenzyme A; CIT, citrate; ICIT, isocitrate; GOX, glyoxylate; AKG, α-ketoglutaric acid; SUCCoA, succinyl-CoA; SUC, succinic acid; FUM, fumaric acid; MAL, malic acid; OAA, oxaloacetic acid; Gly, glycine; 5-ALA, 5-aminolevulinic acid; Amp: ampicillin; Cr: chloramphenicol; ori, replicon. d Specific enzyme activity of wild-type AGXT, three library strains, AGXT22, and AGXT26. e 5-aminolevulinic acid concentrations of MG1655K, MG1655-HAA, MG1655-HAA22, and MG1655-HAA26. All the data are the average values of three independent experiments.**P < 0.01

Fig. 7

Crystal structures of wild-type AGXT and mutants. a PLP binding residues in a single subunit. K209 is both a catalytic residue and one of the PLP-binding residues. The catalytic lysine and PLP-binding residues are shown as magenta sticks. b Polypeptide binding sites of a single subunit. The polypeptide binding residues are shown as cyan sticks. c Mutation sites in AGXT26. The 13 wild-type residues are marked on the left, and the corresponding mutant residues in AGXT26 are marked on the right. Wild-type residues were shown as blue sticks, the mutant residues are shown as red sticks; PLP (pyridoxal-5’-phosphate) is shown in yellow. The image was rendered using PyMOL. d Details of the mutation sites in AGXT26. ^a Synonymous mutation are underlined